Our research on urban scaling, published in Science Advances, calls into question an influential theory of the self-reinforcing dynamics of city growth. We show that big cities feed on their hinterlands to sustain growth, thereby escalating the increasingly uneven economic geography observed in many countries with growing levels of inequality between urban and rural areas.

Cities’ attraction of talent escalates the regional divide.Cities’ attraction of talent escalates the regional divide. Photo credit: Frans Willemblok

Migration of talented people

An influential research field, known as urban scaling, demonstrates that cities’ wealth, innovation, and social change follow highly predictable patterns grounded in population size. Quantifying these agglomeration effects revealed that many socioeconomic indicators increase not simply proportionally to city size but are subject to nonlinear dynamics: Doubling city size, for example, raises total income, the number of patents, the number of residential moves, and the number of romantic breakups by roughly 115%—or 15% per capita—suggesting that urbanites’ productivity and their pace of life increases as their cities grow.

The higher than expected productivity of larger cities reflects differences in cities’ sociodemographic composition. These differences amplify through cities’ attraction of talent from their hinterlands. In Sweden, the better educated and more intelligent are most likely to leave rural areas and move to the largest regions Stockholm (in the country’s east), Gothenburg, or Malmö (both in the south). The gray ties indicate migration flows from smaller to relatively larger labor market areas in 2012.


The higher than expected outputs of larger cities have been explained as an emergent consequence of increased social interactions in dense urban environments. We use Swedish register data with unique granularity to ask how much of this overproportional or “superlinear” urban growth can truly be attributed to increased social interconnectivity. Our geocoded microdata—including individuals’ cognitive ability, educational attainment, professional experience, and wage earnings—capture the dissimilarities in the population composition between metropolitan areas of different size and provide better information on the micromechanisms responsible for observed nonlinearities.

Our results demonstrate that superlinear scaling relations substantively reflect differences in cities’ sociodemographic composition and that those differences are generated by selective migration of highly productive individuals into larger cities. Social interactions explain only half of the previously reported agglomeration effects and, in contrast to existing explanations of urban scaling, differences in population characteristics between metropolitan areas crucially drive the +15% phenomenon.

These results are of considerable policy relevance, because we identify the migration of talented people from smaller areas to larger cities as an important force behind observed agglomeration effects. Those moving from smaller areas to one of Sweden’s larger labor markets have, on average, 1.8 more years of education and their cognitive ability (measured for men in a standardized test during military conscription) is 0.4 standard deviations higher than for those who stayed. This is consequential for societies because strongly selective migration has cumulative effects on local populations in both sending and receiving regions. Big cities feed on their hinterlands to sustain growth, escalating the urban-rural divide in economic prosperity and individual life chances. This study identifies an important mechanism behind the increasingly uneven economic geography observed in many countries in which cities’ attraction of talent adds to growing levels of inequality between urban and rural areas.

This research has been published in: Keuschnigg, Mutgan, Hedström, Sci. Adv. 2019;5: eaav0042